diff mbox series

[31/34] drm/amdkfd: add svm range validate timestamp

Message ID 20210415012337.1755-32-Felix.Kuehling@amd.com (mailing list archive)
State New, archived
Headers show
Series Add HMM-based SVM memory manager to KFD v5 | expand

Commit Message

Felix Kuehling April 15, 2021, 1:23 a.m. UTC
With xnack on, add validate timestamp in order to handle GPU vm fault
from multiple GPUs.

If GPU retry fault need migrate the range to the best restore location,
use range validate timestamp to record system timestamp after range is
restored to update GPU page table.

Because multiple pages of same range have multiple retry fault, define
AMDGPU_SVM_RANGE_RETRY_FAULT_PENDING to the long time period that
pending retry fault may still comes after page table update, to skip
duplicate retry fault of same range.

If difference between system timestamp and range last validate timestamp
is bigger than AMDGPU_SVM_RANGE_RETRY_FAULT_PENDING, that means the
retry fault is from another GPU, then continue to handle retry fault
recover.

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 17 +++++++++++++++++
 drivers/gpu/drm/amd/amdkfd/kfd_svm.h |  2 ++
 2 files changed, 19 insertions(+)
diff mbox series

Patch

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index 40e1dc42a981..7b698b0e4e41 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -35,6 +35,11 @@ 
 
 #define AMDGPU_SVM_RANGE_RESTORE_DELAY_MS 1
 
+/* Long enough to ensure no retry fault comes after svm range is restored and
+ * page table is updated.
+ */
+#define AMDGPU_SVM_RANGE_RETRY_FAULT_PENDING	2000
+
 static void svm_range_evict_svm_bo_worker(struct work_struct *work);
 static bool
 svm_range_cpu_invalidate_pagetables(struct mmu_interval_notifier *mni,
@@ -269,6 +274,7 @@  svm_range *svm_range_new(struct svm_range_list *svms, uint64_t start,
 	INIT_LIST_HEAD(&prange->deferred_list);
 	INIT_LIST_HEAD(&prange->child_list);
 	atomic_set(&prange->invalid, 0);
+	prange->validate_timestamp = ktime_to_us(ktime_get());
 	mutex_init(&prange->migrate_mutex);
 	mutex_init(&prange->lock);
 	svm_range_set_default_attributes(&prange->preferred_loc,
@@ -1392,6 +1398,9 @@  static int svm_range_validate_and_map(struct mm_struct *mm,
 unreserve_out:
 	svm_range_unreserve_bos(&ctx);
 
+	if (!r)
+		prange->validate_timestamp = ktime_to_us(ktime_get());
+
 	return r;
 }
 
@@ -2131,6 +2140,7 @@  svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
 	struct svm_range_list *svms;
 	struct svm_range *prange;
 	struct kfd_process *p;
+	uint64_t timestamp;
 	int32_t best_loc, gpuidx;
 	int r = 0;
 
@@ -2166,6 +2176,13 @@  svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
 	}
 
 	mutex_lock(&prange->migrate_mutex);
+	timestamp = ktime_to_us(ktime_get()) - prange->validate_timestamp;
+	/* skip duplicate vm fault on different pages of same range */
+	if (timestamp < AMDGPU_SVM_RANGE_RETRY_FAULT_PENDING) {
+		pr_debug("svms 0x%p [0x%lx %lx] already restored\n",
+			 svms, prange->start, prange->last);
+		goto out_unlock_range;
+	}
 
 	best_loc = svm_range_best_restore_location(prange, adev, &gpuidx);
 	if (best_loc == -1) {
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
index 11b04a706ff9..aa829b3c992d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h
@@ -86,6 +86,7 @@  struct svm_work_list_item {
  * @actual_loc: the actual location, 0 for CPU, or GPU id
  * @granularity:migration granularity, log2 num pages
  * @invalid:    not 0 means cpu page table is invalidated
+ * @validate_timestamp: system timestamp when range is validated
  * @notifier:   register mmu interval notifier
  * @work_item:  deferred work item information
  * @deferred_list: list header used to add range to deferred list
@@ -122,6 +123,7 @@  struct svm_range {
 	uint32_t			actual_loc;
 	uint8_t				granularity;
 	atomic_t			invalid;
+	uint64_t			validate_timestamp;
 	struct mmu_interval_notifier	notifier;
 	struct svm_work_list_item	work_item;
 	struct list_head		deferred_list;